Tech debt: CapEx or OpEx? IRC §174 implications

The Pre-2022 World

How R&D was treated before TCJA effective date

Before the 2022 tax year, Internal Revenue Code §174 allowed taxpayers to elect immediate deduction of research and experimental expenditures in the year incurred. This treatment had been the norm since 1954 and was a structural feature of how technology companies operated in the US tax system. R&D spend reduced current-year taxable income on a dollar-for-dollar basis, regardless of whether the underlying expenditure was funded from operating cash or from invested capital.

For a software company with $10M of engineering payroll, the pre-2022 treatment generated a $10M deduction in the current year, which at a 21% federal corporate rate reduced cash taxes by $2.1M (state taxes added another 5-10% on top). The deduction matched the cash outflow, so engineering investment was tax-neutral from a cash-timing perspective.

The Tax Cuts and Jobs Act of 2017, signed into law in December 2017, included a delayed provision modifying §174. The change was scheduled to take effect with the 2022 tax year. The change was structural and was widely expected to be reversed before it took effect; it was not, and the new regime has been the operating reality since 2022 with no legislative reversal despite multiple proposals.

The Current Treatment

What §174 requires now

Under the post-TCJA §174, specified research and experimental expenditures must be capitalized and amortized over a defined period, with a half-year convention in the first year. The amortization period depends on where the R&D is performed.

See Rev. Proc. 2023-08 for the procedural guidance on implementing §174 changes, and Notice 2023-63 for additional interim guidance on the scope of specified R&D expenditures.

A $10M engineering payroll spend that previously generated a $10M deduction in year one now generates a $1M deduction in year one and $2M deductions in each of years two through five, with the remaining $1M deducted in year six (the half-year convention defers some of the deduction to the year after the amortization period ends). The total deduction over the amortization period is the same; the timing is materially different. The cash-tax impact depends on the company's tax position; for profitable companies in the 21% federal rate the cash impact is $1.9M of additional cash taxes in year one of the work.

The Scope Question

What counts as §174-eligible R&D

The scope of §174-eligible expenditures is broader than many CFOs assume. The statutory language captures expenditures “in connection with the taxpayer's trade or business which represent research and development costs in the experimental or laboratory sense.” Treasury and IRS interpretive guidance has extended this to most software-development activity, including software for internal use, software for sale or license, and (importantly for this discussion) significant refactoring or improvement of existing software.

The interpretive distinction is between “maintenance” (routine bug fixes, minor patches, ongoing support) and “development” (changes that improve capability, performance, or functionality). Maintenance work generally falls outside §174; development work generally falls inside. The line is judgement-based and tax counsel involvement is appropriate for material amounts. For most tech-debt remediation work, the answer is that the work falls inside §174 because the remediation typically improves capability (better testability, better performance, better maintainability that enables future development).

A CFO who wants to optimise the §174 treatment can structure engineering work to maximise the maintenance portion (which gets immediate expensing) and minimise the §174 capture, but the optimisation has limits because Treasury guidance is broad and aggressive maintenance-classification can attract IRS scrutiny. The pragmatic approach for most companies is to accept the §174 treatment as the planning baseline and to model the cash-tax impact when evaluating tech-debt investment decisions.

The GAAP-Tax Distinction

Why financial statements and tax returns disagree

GAAP financial reporting and tax accounting are separate systems with different rules. GAAP treatment of software development is governed by ASC 350-40 (internal-use software, which generally requires capitalization of certain post-preliminary-stage development costs) and ASC 985-20 (software for sale or license, which has its own capitalization rules tied to technological feasibility). These rules predate the §174 changes and are independent of them.

The result is that a single engineering expenditure can be treated three different ways in three different systems: immediately expensed under GAAP for R&D opex purposes (most non-internal-use software development), capitalized over a different amortization period under GAAP for capitalizable software, and capitalized over five or fifteen years under §174 for tax purposes. The differences create deferred-tax positions that the CFO and external auditor track jointly, and that show up on the balance sheet as deferred tax assets or liabilities.

For tech-debt remediation specifically, the most common pattern is: immediate expensing under GAAP (because the work is not creating a new capitalizable asset, just maintaining or improving an existing one), §174 capitalization for tax (because the work is §174-eligible R&D). The result is a deferred tax asset (the company has paid taxes on income that has not yet been deducted; the deduction will come in future years as the §174 amortization unwinds). This is structurally similar to how depreciation timing differences are treated, and the CFO's tax planning is similar in principle.

The CFO Calculation

Modelling §174 into tech-debt funding decisions

The CFO's tech-debt funding decision now includes a §174 cash-tax timing adjustment that did not exist before 2022. A $2M tech-debt initiative previously generated an immediate $420K cash-tax benefit (at 21% federal); under §174 the year-one cash-tax benefit is $42K, with the remaining $378K spread across years 2-6. The net present value of the deferred deduction depends on the discount rate, but at a 10% cost of capital the deferred-deduction NPV is roughly $300K rather than $420K, a $120K effective additional cost of the initiative.

For tech-debt initiatives where the payback is measured in months, the §174 timing impact is small relative to the operational savings. For larger structural initiatives where the payback is measured in years, the §174 impact is material and should be modelled into the decision. A CTO who pitches an initiative without modelling the §174 impact may find the CFO declining on grounds the CTO does not understand; learning the §174 framework is a useful complement to the cost-of-inaction modelling covered on the CFO pitch page.

The §174 framework is particularly important for early-stage and growth-stage companies whose tax positions are evolving rapidly. A company that was loss-making (and therefore not benefiting from §174 deductions at all) and that crosses into profitability faces a sudden visible impact from §174 amortization timing on the year-one tax bill. The CFO's tax planning should anticipate this transition, and the CTO's tech-debt funding strategy should reflect the changing cash-tax dynamics.

Cross-Reference

CapEx vs OpEx in the financial-framework stack

CapEx vs OpEx is one of four financial-framework treatments. The companion pages: debt service ratio (borrowing the consumer-credit metaphor), technical insolvency (the velocity-to-zero inflection point), and enterprise value at exit (the M&A and IPO valuation impact). For the stakeholder framing that uses the §174 treatment most directly, see the CFO pitch and the public company stage page.

The COGS-vs-opex GAAP question covered on the COGS page is related but separate. COGS classification is a GAAP financial-reporting decision; §174 is a tax accounting decision. The two can apply to the same expenditure independently.

Field Notes

Frequently asked questions

Adjacent Reading